Analyze ingested GA4 data using platform query tools
Join a GA4-sourced event dataset with a loyalty profile dataset on a shared key, build an analysis workspace with attribution-aware freeform tables and funnel visualizations, and derive actionable conversion insights from the combined data.
Once GA4 event data has landed in the platform, the analytical value is realized by joining it with first-party profile data — loyalty membership, purchase history, identity graph — and then running funnel, flow, and attribution analyses that were impossible when the data lived solely in a Google Analytics report. This task covers the configuration and workspace-assembly steps required to turn the raw ingested events into business intelligence.
The join configuration is the highest-stakes decision: selecting loyaltyId as the Person ID links anonymous GA4 sessions to known customers only for sessions where the user was authenticated. Analysts must decide how to handle unauthenticated sessions — whether to exclude them, group them under a device ID, or apply probabilistic stitching. Once the connection is live, workspace analyses benefit from the enriched schema: marketing channel attribution can be evaluated across both web and in-store touchpoints, and loyalty tier can be used as a filter to compare conversion funnels for high-value versus standard members.
Parallel viability: High parallelism. The same join and analysis pattern is reproducible in Looker (with BigQuery as the backend), Tableau (with Snowflake or BigQuery as the source), or any BI tool that can join a GA4 export table to a CRM/loyalty table. The AEP-specific advantage is that the joined data also feeds into Real-time Customer Profile for activation use cases (audience building, journey triggering), creating a closed loop that a pure BI tool cannot replicate. Teams whose GA4 analysis is purely retrospective and does not feed real-time activation should evaluate whether a composable BI stack meets their needs at lower cost.
Side-by-side implementations
After ingesting BigQuery GA4 event data into AEP, practitioners create a CJA Connection that includes both the BigQuery-sourced dataset and the AEP Loyalty Profile dataset, joining them on loyaltyId as the Person ID. A Data View is built on the connection exposing GA4 dimensions (Marketing Channel, Device Type, Page names) and metrics. Analysis Workspace projects are then assembled: freeform tables with column-level attribution model overrides (e.g., Last Touch vs. Linear for Marketing Channel), fallout visualizations tracking the e-commerce funnel (Home → Store Search → Product_Detail_Views → Product_Checkouts), and flow analyses using Marketing Channel as the dimension to reveal channel-switching paths. The combined GA4 + loyalty data enables segmentation by loyalty tier within the same workspace.
Capability: Identity Resolution
In Snowflake, the equivalent of CJA's Connection joining GA4 and Loyalty Profile datasets is a dbt model that JOINs the flattened GA4 events table to the CRM/loyalty profile table on a shared identifier (loyaltyId or hashed email). The Analysis Workspace fallout funnel (Home → Store Search → Product Detail → Checkout) is a Tableau or Looker funnel visualization built on a Snowflake SQL query using sequential event filtering with CONDITIONAL_CHANGE_EVENT or window-function patterns. Flow analysis is implemented as a Sankey chart in Tableau or a Looker block. Attribution model overrides at the column level are configured in the BI tool's calculation editor. The person-level join is the composable equivalent of CJA's cross-channel stitching — the join key is the person_id column from the shared identity map table rather than AEP's Identity Service.
Capability: Identity Resolution
Hightouch is not an analytics or BI tool and has no equivalent to CJA's Analysis Workspace, fallout funnels, or flow visualizations. Ad-hoc GA4 analysis and funnel/attribution reporting remain entirely in the Snowflake + dbt + BI layer documented in Phase 3. Hightouch's role in the post-analysis workflow is activation: once Snowflake/BI analysis identifies a GA4-derived cohort (e.g., "users who viewed 3+ product pages but did not add to cart"), that cohort is expressed as a Hightouch Audience filter against the dbt GA4 model and activated to a marketing destination in real time. This closes the analytics-to-activation loop that CJA + AEP provides natively via the shared data layer: the BI tool produces the insight, Hightouch translates it into an audience, and the audience flows to a downstream channel without requiring data re-export or manual list uploads.
Capability: Audience Segmentation
Task-level sources
- technical-training/module12/index.md
- technical-training/module12/ex5.md
- technical-training/module12/summary.md
How is this implementation?
Sign-in-gated. Tomorrow morning's curriculum-ingestor consumes your feedback: "Inaccurate" queues the task for re-review, "needs update" queues it for a refresh, and "one vendor panel is wrong" re-drafts just that panel.